## 12. Revenue Performance Index (Income-Based Classification)

## Source details

**Canonical URL:** [12. Revenue Performance Index (Income-Based Classification)](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp07184.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp07184.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp07184.pdf.json)

---

### I. Introduction and Objective
- Objective: Investigate main determinants of central government revenue (excluding grants) as a share of GDP across developing countries, using an extended dataset and econometric corrections for persistence and endogeneity.
- Data scope: 105 developing countries over 25 years.
- Main extensions relative to prior literature:
  - Larger country sample and time span.
  - New explanatory variables: specific tax sources, political stability, economic stability, law and order.
  - Econometric specifications addressing persistence of revenue performance and potential feedback of revenues on explanatory variables.
- Principal summary findings:
  - Structural factors — per capita GDP, share of agriculture in GDP, and trade openness — are strong determinants of revenue performance.
  - Foreign aid improves revenue performance; foreign debt does not have a robust positive effect.
  - Corruption is a significant adverse institutional determinant of revenue performance.
  - Political and economic stability matter in some specifications but effects are not robust across all specifications.
  - Countries that rely primarily on taxes on goods and services have relatively poor revenue performance; countries relying more on income taxes, profit taxes, and capital gains taxes perform better.
- Constructed revenue performance index: actual revenue / predicted revenue to compare observed performance against model-based potential.

### II. Data Description (selected exact figures preserved)
- Key data sources:
  - Central government revenue (% of GDP): GFS & WETA.
  - Per capita GDP (PPP), sector shares, imports (% of GDP), aid (% of GNI): WDI.
  - Debt (% of GNI): IFS.
  - Tax source composition: GFS.
  - Institutional indicators: ICRG.
  - Trade restrictiveness / average tariff: IMF.
- Sample summary statistics (as reported in Table 1):
  - Central government revenue (% of GDP): No. of Obs. 2,013; Percentage Available 67; Mean 19.8; Std. Dev 13.2; Min -225; Max 79.33.
  - Per capita GDP in PPP: No. of Obs. 2,587; Percentage Available 86.2; Mean 8.2; Std. Dev 0.96; Min 1; Max 10.72.
  - Agriculture, value added (% of GDP): No. of Obs. 2,448; Percentage Available 81.6; Mean 21.8; Std. Dev 14.5; Min 0; Max 72.03.
  - Imports (% of GDP): No. of Obs. 2,551; Percentage Available 85.0; Mean 43.0; Std. Dev 22.8; Min 1; Max 173.48.
  - Aid (% of GNI): No. of Obs. 2,562; Percentage Available 85.4; Mean 8.6; Std. Dev 13.4; Min -1210.56; Max 56.
  - Debt (% of GNI): No. of Obs. 2,277; Percentage Available 75.9; Mean 5.8; Std. Dev 4.9; Min 0; Max 80.76.
  - Tax revenue from goods and services (% of total revenue): No. of Obs. 756; Percentage Available 25.2; Mean 28.3; Std. Dev 15.3; Min 0; Max 76.74.
  - Tax revenue from income, profits and capital gains (% of total revenue): No. of Obs. 736; Percentage Available 24.5; Mean 20.6; Std. Dev 12.9; Min 0; Max 79.54.
  - Tax revenue from trade (% of total revenue): No. of Obs. 747; Percentage Available 24.9; Mean 16.5; Std. Dev 14.2; Min 0; Max 64.66.
  - Political stability (ICRG): No. of Obs. 1,711; Percentage Available 57.0; Mean 5.7; Std. Dev 7.7; Min 13.7; Max 99.0.

### III. Empirical Strategy and Estimation Techniques
- Estimation approaches:
  - Fixed effects and random effects panel regressions (baseline).
  - Panel-corrected standard error (PCSE) estimates (Prais-Winsten) allowing for AR(1) processes and contemporaneous correlation.
  - Dynamic panel estimators: difference-GMM and system-GMM (Arellano-Bond; Arellano-Bover / Blundell-Bond).
  - Sensitivity checks: lagged aid and debt to test endogeneity; subsample analysis by World Bank income classification (low-, middle-, high-income).
- Endogeneity and persistence:
  - Lagged dependent variable significant in dynamic panel estimations documenting persistence.
  - System-GMM results broadly similar to PCSE preferred specifications once persistence is addressed.
  - Re-estimations using lagged values of aid and debt indicate endogeneity is not a severe problem; aid remains positively and significantly associated with revenue performance.

### IV. Key Empirical Results (preserved magnitudes and phrasing)
- Per capita GDP (log):
  - Positive and significant in most random-effects regressions and many fixed-effects specifications.
  - PCSE and system-GMM confirm positive association; impact smaller in dynamic (system-GMM) specifications.
- Agriculture share:
  - Strong negative and significant relationship with revenue performance.
  - Example magnitude: "a one percent increase in the share of agriculture sector could reduce revenue performance by as much as 0.4 percent."
  - Significant across income groups.
- Import share (trade openness):
  - Generally positive and often significant.
  - Example magnitude: "an increase in imports/GDP of one percent may increase revenue performance by up to 0.15 percent."
- Foreign aid:
  - Positive effect on revenue performance; effect stronger in PCSE and system-GMM where autocorrelation is accounted for.
  - Example projection: "an anticipated increase in aid from around US$80 billion in 2004 to US$130 billion in 2010 would increase revenue performance by as much as 0.6 percent."
  - Aid effect particularly significant for low-income countries.
- Foreign debt:
  - Relationship not robustly positive; in some specifications debt is negatively related to revenue performance, but effect is generally weak.
- Institutional factors:
  - Corruption: consistently significant adverse effect in preferred specifications (PCSE with panel specific correlation coefficient and some GMM specifications).
  - Political and economic stability: significant in some specifications and subsamples (not always robust).
  - Government stability, law and order: generally not significant across most specifications.
- Tax base composition:
  - Greater reliance on taxes on goods and services (indirect taxes) associated with lower revenue performance (negative and significant coefficients in many specifications).
  - Greater reliance on taxation of income, profits and capital gains associated with higher revenue performance (positive and significant coefficients).
  - Share of tax revenue from trade generally not significant or not robust.
- Tax rates and average tariffs:
  - Highest marginal corporate and individual tax rates and average tariffs do not significantly explain revenue performance once structural, institutional and tax composition variables are included; these variables were dropped from subsequent analysis.

### V. Subsample (Income-Group) Findings (selected exact reported magnitudes)
- Low-income countries:
  - Agriculture share negative and significant.
  - Import share positive and significant.
  - Foreign aid significant positive: "an increase in foreign aid by 1 percent can improve revenue performance by as much as 0.11 percent."
  - Corruption significant: "in low-income countries, an increase in the corruption index of one unit would improve revenue performance by about 1.5 percent."
  - Political stability: weakly positive (example: one unit increase can increase revenue performance by 0.08 percent).
  - Tax composition: taxing goods and services negatively related; income/profit/capital gains positive.
- Middle-income countries:
  - Agriculture share negative and significant.
  - Per capita GDP effect weaker than in high-income countries.
  - Corruption significant and negative; political stability weakly positive.
  - Tax composition patterns similar to other groups.
- High-income countries:
  - Per capita GDP has strong impact.
  - Agriculture share negative and significant.
  - Taxing goods and services negatively associated with revenue performance; income/profit/capital gains positive.

### VI. Revenue Effort Index: Construction, Robustness, and Patterns
- Method:
  - Predicted revenue performance computed from regression coefficients (preferred specifications cited: column (III) and (IX) of Table 5).
  - Revenue effort index = actual revenue performance / predicted revenue performance.
  - Index interpretation: =1 observed equals predicted; >1 above predicted; <1 below predicted.
- Robustness:
  - Two model variants for predicted values: one including per capita GDP and one including agriculture share (due to collinearity).
  - Correlation between indices from the two specifications: R^2 = 0.76.
  - Differences typically <0.3 and often <0.1.
- Key index patterns (examples preserved):
  - Using full sample specification including per capita GDP, 43 countries perform better than predicted (42 when using agriculture share specification).
  - Notable over-performers (examples): Burundi, Botswana, Malawi, Zimbabwe, Ethiopia, Guinea-Bissau.
  - Notable under-performers (examples): Argentina, Brazil, Peru, Panama, United Arab Emirates.
- Tabulations:
  - Table 11 and Table 12 present revenue effort indices for developing countries (1980–2004) and by income group under Per Capita GDP and Agriculture Share specifications.

### VII. Policy Implications and Recommendations (preserved substance)
- Main conclusions:
  - Principal determinants of revenue performance: per capita GDP, agriculture’s share in GDP, trade openness, foreign aid, corruption, political stability (in some specifications), and tax composition.
  - Foreign aid has a positive impact on revenue performance, especially for low-income countries.
  - Corruption has a significant negative effect on revenue performance, notably for low- and middle-income countries.
  - Reliance on indirect taxes (goods and services) tends to be associated with weaker revenue performance; reliance on direct taxes (income, profits, capital gains) is associated with stronger revenue performance.
  - Geographic patterns: Sub-Saharan African countries include both strong over-performers and under-performers; many Latin American and Eastern European countries fall short of their revenue potential.
- Explicit policy recommendations:
  - Increase aid targeted to low-income countries, given the positive impact of foreign aid on revenue performance in this group: "The positive impact of foreign aid on revenue performance, especially for low-income countries, recommends increased aid to these countries."
  - Strengthen anti-corruption measures and improve institutional quality to enhance revenue mobilization.
  - Consider tax policy reforms that broaden the base and increase reliance on progressive direct taxes (income, profits, capital gains) rather than over-reliance on indirect taxes on goods and services.
  - Improve trade openness and customs procedures to harness the positive relationship between trade openness (imports share) and revenue performance.
  - Recognize structural constraints (low per capita GDP, large agriculture shares) when setting realistic revenue targets; use revenue effort indices to identify countries that can realistically mobilize more domestic revenue and those that face binding structural limits.

### VIII. Aid Commitments, VAT, Corruption, and Tax Structure (selected exact figures and phrases)
- Aid commitments and actual flows:
  - Donor benchmark: "0.7 percent of their GNP in international aid."
  - Actual flow in 2003: "total aid from the 22 richest countries to the world's developing countries was just US$69 billion—a shortfall of US$130 billion from the 0.7 percent promise."
  - Average provision by richest countries: "just 0.25 percent of their GNP in official development assistance."
  - Policy implication: donor countries should "monitor the aid flow and ensure that it is used for poverty reduction and infrastructure development, which would generate higher revenue in the future."
- Corruption and political stability:
  - Reduction in corruption and increased political stability are "expected to improve revenue performance of low-income and middle- income countries."
  - Recommendations: "reduce the opportunities for corruption in tax administration" and "change the incentive structure for tax officials."
  - Political instability consequences: governments "face a credibility problem" preventing long-term investments and lowering tax revenue.
  - "The low-income countries would also benefit from a stable political regime."
- Tax structure and VAT considerations:
  - Empirical relations: "positive relation between taxes and revenue performance" and "negative relation between indirect taxes and revenue performance."
  - Practical constraints: "In most developing countries there are severe problems in raising tax revenue through direct taxes." and difficulty in developing mass personal income tax systems due to poverty.
  - VAT observations: "The self enforcing mechanism of VAT can induce greater compliance." and "By including services in its fold, VAT broadens the tax base and it eliminates the cascading effects involved in turnover taxes and some sales tax systems."
  - Conclusion: "VAT has a greater potential in improving the revenue performance in developing countries, compared to traditional commodity taxes, for a number of reasons."

*Author's calculations and analysis contained in IMF working paper chapter 12.*

### References .............................................................................................................

### _wp07184 - References .............................................................................................................

### Figures
- 1.   Central Government Revenue and Agriculture................................................................... 7
- 2.   Central Government Revenue and Manufacturing ............................................................. 8
- 3.   Central Government Revenue and Log of Per Capita GDP................................................ 8
- 4.   Central Government Revenue and Imports......................................................................... 9
- 5.   Central Government Revenue and Political Stability ......................................................... 9
- 6.   Central Government Revenue and Economic Stability .................................................... 10
- 7.   Variation in Revenue Performance ................................................................................... 15

### Tables
- 1.   Summary of Variables.........................................................................................................6
- 2.   Determinants of Revenue Performance (Fixed Effects Estimation) ................................. 13
- 3.   Determinants of Revenue Performance (Random Effects Estimation)............................. 14
- 4.   Determinants of Revenue Performance (Common Correlation Coefficient).................... 17
- 5.   Determinants of Revenue Performance (Panel Specific Correlation Coefficient)............ 18
- 6.   Determinants of Revenue Performance (Lagged Values of Foreign Aid and Debt) ........ 19
- 7.   Determinants of Revenue Performance (Dynamic Panel Specification) .......................... 21
- 8.   Determinants of Revenue Performance (Low-Income Countries).................................... 23
- 9.   Determinants of Revenue Performance (Middle-Income Countries) ............................... 24
- 10. Determinants of Revenue Performance (High-Income Countries) ................................... 25
- 11. Revenue Effort Indices for Developing Countries (1980–2004) ...................................... 28

*Source: _wp07184 - References .............................................................................................................*

### 12. Revenue Performance Index (Income-Based Classification) ........................................... 29

### 12. Revenue Performance Index (Income-Based Classification)

### I. Introduction
- Objective: Investigate main determinants of central government revenue (excluding grants) as a share of GDP across developing countries, using an extended dataset and econometric corrections for persistence and endogeneity.
- Data scope: 105 developing countries over 25 years.
- Main extension relative to prior literature: larger country sample and time span; incorporation of new explanatory variables (specific tax sources, political stability, economic stability, law and order); econometric specifications addressing persistence of revenue performance and potential feedback of revenues on explanatory variables.
- Principal summary findings:
  - Structural factors — per capita GDP, share of agriculture in GDP, and trade openness — are strong determinants of revenue performance.
  - Foreign aid improves revenue performance; foreign debt does not have a robust positive effect.
  - Corruption is a significant adverse institutional determinant of revenue performance.
  - Political and economic stability matter in some specifications but effects are not robust across all specifications.
  - Countries that rely primarily on taxes on goods and services have relatively poor revenue performance; countries relying more on income taxes, profit taxes, and capital gains taxes perform better.
- Constructed a revenue performance index (actual revenue / predicted revenue) to compare observed performance against model-based potential.

### II. Recent Research Findings (Literature synthesis)
- Common explanatory variables in prior studies: per capita GDP, sectoral composition (agriculture, mining), trade openness, aid/GDP, debt/GDP, size of informal economy, institutional measures (political stability, corruption).
- Typical empirical regularities reported in the literature:
  - Per capita GDP and trade openness generally positively correlated with tax shares.
  - Higher agriculture share typically associated with lower tax shares.
  - Mining/resource shares show ambiguous relationships across studies.
  - Mixed findings on foreign debt and aid; effects sensitive to country sample and period.
- Noted contributions in literature: Chelliah (1971), Lotz and Morss (1967), Tanzi (1992), Ghura (1998), Bird, Martinez-Vasquez and Torgler (2004), Keen and Simone (2004), Rodrik (1998).

### III. Data Description
- Dataset coverage and key variable sources:
  - Central government revenue (% of GDP): GFS & WETA.
  - Structural variables: per capita GDP (PPP) and sector shares from WDI; imports (% of GDP) from WDI; aid (% of GNI) from WDI; debt (% of GNI) from IFS.
  - Tax source composition (goods & services; income, profits and capital gains; trade; exports) from GFS.
  - Highest marginal tax rates (individual and corporate) from WDI.
  - Institutional indicators (political stability, economic stability, corruption, law and order, government stability): ICRG.
  - Trade restrictiveness / average tariff: IMF.
- Table 1 summary statistics (selected exact figures preserved as in source):
  - Central government revenue (% of GDP): No. of Obs. 2,013; Percentage Available 67; Mean 19.8; Std. Dev 13.2; Min -225; Max 79.33.
  - Per capita GDP in PPP: No. of Obs. 2,587; Percentage Available 86.2; Mean 8.2; Std. Dev 0.96; Min 1; Max 10.72.
  - Agriculture, value added (% of GDP): No. of Obs. 2,448; Percentage Available 81.6; Mean 21.8; Std. Dev 14.5; Min 0; Max 72.03.
  - Imports (% of GDP): No. of Obs. 2,551; Percentage Available 85.0; Mean 43.0; Std. Dev 22.8; Min 1; Max 173.48.
  - Aid (% of GNI): No. of Obs. 2,562; Percentage Available 85.4; Mean 8.6; Std. Dev 13.4; Min -1210.56; Max 56.
  - Debt (% of GNI): No. of Obs. 2,277; Percentage Available 75.9; Mean 5.8; Std. Dev 4.9; Min 0; Max 80.76.
  - Tax revenue from goods and services (% of total revenue): No. of Obs. 756; Percentage Available 25.2; Mean 28.3; Std. Dev 15.3; Min 0; Max 76.74.
  - Tax revenue from income, profits and capital gains (% of total revenue): No. of Obs. 736; Percentage Available 24.5; Mean 20.6; Std. Dev 12.9; Min 0; Max 79.54.
  - Tax revenue from trade (% of total revenue): No. of Obs. 747; Percentage Available 24.9; Mean 16.5; Std. Dev 14.2; Min 0; Max 64.66.
  - Political stability (ICRG): No. of Obs. 1,711; Percentage Available 57.0; Mean 5.7; Std. Dev 7.7; Min 13.7; Max 99.0 (note: table entries preserved as in source).

### IV. Empirical Analysis and Key Results
- Estimation approaches deployed:
  - Fixed effects and random effects panel regressions (baseline).
  - Panel-corrected standard error (PCSE) estimates (Prais-Winsten) allowing for AR(1) processes and contemporaneous correlation.
  - Dynamic panel estimators: difference-GMM and system-GMM (Arellano-Bond; Arellano-Bover / Blundell-Bond).
  - Sensitivity checks: lagged aid and debt to test endogeneity; subsample analysis by World Bank income classification (low-, middle-, high-income).
- Robust empirical regularities (preserved coefficients and qualitative magnitudes as reported):
  - Log of per capita GDP:
    - Positive and significant in most random-effects regressions and many fixed-effects specifications.
    - PCSE and system-GMM confirm positive association; impact smaller in dynamic (system-GMM) specifications.
  - Agriculture share:
    - Strong negative and significant relationship with revenue performance.
    - Example magnitude reported: "a one percent increase in the share of agriculture sector could reduce revenue performance by as much as 0.4 percent" (source statement preserved).
    - Significant across income groups.
  - Import share (trade openness):
    - Generally positive and often significant; example magnitude: an increase in imports/GDP of one percent may increase revenue performance by up to 0.15 percent.
  - Foreign aid:
    - Positive effect on revenue performance; effect stronger in PCSE and system-GMM where autocorrelation is accounted for.
    - Example projection: an anticipated increase in aid from around US$80 billion in 2004 to US$130 billion in 2010 would increase revenue performance by as much as 0.6 percent (exact phrasing preserved).
    - Aid effect particularly significant for low-income countries.
  - Foreign debt:
    - Relationship not robustly positive; in some specifications debt is negatively related to revenue performance, but effect is generally weak.
  - Institutional factors:
    - Corruption: consistently significant adverse effect in preferred specifications (PCSE with panel specific correlation coefficient and some GMM specifications).
    - Political and economic stability: significant in some specifications and subsamples (not always robust).
    - Government stability, law and order: generally not significant across most specifications.
  - Tax base composition:
    - Greater reliance on taxes on goods and services (indirect taxes) associated with lower revenue performance (coefficients negative and significant in many specifications).
    - Greater reliance on taxation of income, profits and capital gains associated with higher revenue performance (coefficients positive and significant).
    - Share of tax revenue from trade generally not significant or not robust.
  - Tax rates and average tariffs:
    - Highest marginal corporate and individual tax rates and average tariffs do not significantly explain revenue performance once structural, institutional and tax composition variables are included; these variables were dropped from subsequent analysis.

- Persistence and dynamics:
  - Substantial persistence of revenue performance documented via country time series; lagged dependent variable significant in dynamic panel estimations.
  - System-GMM yields results broadly similar to PCSE preferred specifications once persistence is addressed.

- Sensitivity to endogeneity:
  - Re-estimations using lagged values of aid and debt indicate endogeneity is not a severe problem; aid remains positively and significantly associated with revenue performance.

- Subsample (income-group) findings (preserved exact reported magnitudes where given):
  - Low-income countries:
    - Agriculture share negative and significant.
    - Import share positive and significant.
    - Foreign aid has a significant positive effect; "an increase in foreign aid by 1 percent can improve revenue performance by as much as 0.11 percent."
    - Corruption significant: e.g., "in low-income countries, an increase in the corruption index of one unit would improve revenue performance by about 1.5 percent."
    - Political stability: weakly positive effect (example: one unit increase can increase revenue performance by 0.08 percent).
    - Tax composition: taxing goods and services negatively related in low-income group; income/profit/capital gains positive across groups.
  - Middle-income countries:
    - Agriculture share negative and significant.
    - Per capita GDP effect weaker than in high-income countries.
    - Corruption significant and negative; political stability weakly positive.
    - Tax composition: similar patterns as other groups.
  - High-income countries:
    - Per capita GDP has strong impact.
    - Agriculture share negative and significant.
    - Taxing goods and services negatively associated with revenue performance in high-income group as well; income/profit/capital gains positive.

### V. Assessment of Revenue Performance (Revenue Effort Index)
- Method:
  - Predicted revenue performance computed from regression coefficients (preferred specifications cited: column (III) and (IX) of Table 5).
  - Revenue effort index = actual revenue performance / predicted revenue performance.
  - Index interpretation: value = 1 means observed equals predicted; >1 means above predicted; <1 means below predicted.
- Robustness notes:
  - Two model variants used for predicted values: one including per capita GDP and one including agriculture share (due to collinearity between them). Results largely similar: correlation between indices from the two specifications R^2 = 0.76; differences typically <0.3 and often <0.1.
- Key index patterns reported (selected exact examples preserved):
  - Using the full sample specification including per capita GDP, 43 countries perform better than predicted (dropping to 42 when using agriculture share specification).
  - Notable over-performers (high revenue effort index): Burundi, Botswana, Malawi, Zimbabwe, Ethiopia, Guinea-Bissau (examples preserved in narrative).
  - Notable under-performers (low revenue effort index): Argentina, Brazil, Peru, Panama, United Arab Emirates (examples preserved in narrative).
- Tabulated indices:
  - Table 11 and Table 12 present revenue effort indices by country under Per Capita GDP and Agriculture Share specifications and by income group. (Specific country index values preserved in the source tables.)

### VI. Policy Recommendations and Conclusions
- Main conclusions (preserved exactly in substance):
  - Principal determinants of revenue performance include per capita GDP, agriculture’s share in GDP, trade openness, foreign aid, corruption, political stability (in some specifications), and tax composition.
  - Foreign aid has a positive impact on revenue performance, especially for low-income countries.
  - Corruption has a significant negative effect on revenue performance, notably for low- and middle-income countries.
  - Reliance on indirect taxes (goods and services) tends to be associated with weaker revenue performance; reliance on direct taxes (income, profits, capital gains) is associated with stronger revenue performance.
  - Sub-Saharan African countries include both strong over-performers and under-performers; many Latin American and Eastern European countries fall short of their revenue potential.
- Policy recommendations (as implied and stated in the source):
  - Increase aid targeted to low-income countries, given the positive impact of foreign aid on revenue performance in this group (explicitly referenced in the source: "The positive impact of foreign aid on revenue performance, especially for low-income countries, recommends increased aid to these countries.").
  - Strengthen anti-corruption measures and improve institutional quality to enhance revenue mobilization.
  - Consider tax policy reforms that broaden the base and increase reliance on progressive direct taxes (income, profits, capital gains) rather than over-reliance on indirect taxes on goods and services.
  - Improve trade openness and customs procedures to harness the positive relationship between trade openness (imports share) and revenue performance.
  - Recognize structural constraints (low per capita GDP, large agriculture shares) when setting realistic revenue targets; use revenue effort indices to identify countries that can realistically mobilize more domestic revenue and those that face binding structural limits.

*Author's calculations and analysis contained in IMF working paper chapter 12.*

### 0.7 percent

### 0.7 percent

### Aid commitments and actual flows
- The idea of donors providing "0.7 percent of their GNP in international aid" is described as a potential step in the right direction.
- Actual flow: "In 2003, total aid from the 22 richest countries to the world's developing countries was just US$69 billion—a shortfall of US$130 billion from the 0.7 percent promise."
- Average provision: "On average, the world's richest countries provided just 0.25 percent of their GNP in official development assistance."
- Policy implication highlighted: donor countries should "monitor the aid flow and ensure that it is used for poverty reduction and infrastructure development, which would generate higher revenue in the future."

### Corruption, political stability, and revenue performance
- Reduction in corruption and increased political stability are "expected to improve revenue performance of low-income and middle- income countries."
- Recommendations for developing countries:
  - "reduce the opportunities for corruption in tax administration"
  - "change the incentive structure for tax officials"
- Political instability effects:
  - In politically unstable countries, "the governments face a credibility problem and the government is unable to define and arbitrate property rights."
  - This inability "prevents investors from undertaking long-term investments, which in turn lowers economic growth and overall tax revenue."
- The text stresses that "The low-income countries would also benefit from a stable political regime."

### Tax structure, direct versus indirect taxes, and constraints
- Empirical relations noted:
  - There is a "positive relation between taxes and revenue performance."
  - There is a "negative relation between indirect taxes and revenue performance."
- Despite the above, practical constraints are emphasized:
  - "In most developing countries there are severe problems in raising tax revenue through direct taxes."
  - It is "difficult to develop a mass system of personal income taxes as a significant proportion of the population is extremely poor."
  - In some middle- and high-income developing countries, there is scope "for rationalizing the rate structure and limiting exemptions to improve revenue from personal income tax," noting exemptions "at times amount to several times the country’s per capita GDP and therefore benefit those with high incomes."

### Value-Added Tax (VAT) and indirect taxation
- The traditional argument against most indirect taxes is "its regressivity, which exacerbates inequality and reduces the tax base, which may lead to a reduction in the share of revenue in GDP."
- VAT-specific observations:
  - "In recent years, with the adoption of VAT in many developing countries, the revenue performance response of these have been mixed."
  - VAT advantages listed:
    - "The self enforcing mechanism of VAT can induce greater compliance."
    - "By including services in its fold, VAT broadens the tax base and it eliminates the cascading effects involved in turnover taxes and some sales tax systems."
  - The text concludes: "VAT has a greater potential in improving the revenue performance in developing countries, compared to traditional commodity taxes, for a number of reasons."

### Key statistics and exact figures cited
- "0.7 percent of their GNP in international aid" — stated as the donor commitment benchmark.
- "In 2003, total aid from the 22 richest countries to the world's developing countries was just US$69 billion."
- Shortfall relative to the 0.7 percent promise: "a shortfall of US$130 billion."
- Average aid provision by the world's richest countries: "just 0.25 percent of their GNP in official development assistance."
- Exemption practices: exemptions may "at times amount to several times the country’s per capita GDP" (phrase preserved as in source).

*Source: _wp07184 - 0.7 percent*

### Appendix D.      Summary of Findings of Empirical Studies

### Appendix D.      Summary of Findings of Empirical Studies

### Study summaries
- Lotz and Morss (1967)
  - Significant Variables (Sign): Per capita GNP (+), trade share (+)
  - Other Variables Included in the Regressions: 
    - (none listed)
  - Goodness of Fit: 10 to 60% 
  - Countries Covered: 72 developing countries
  - Time Period: 1962-66

- Chelliah (1971)
  - Significant Variables (Sign): Mining share (+), non mineral export share (+), agriculture share (-)
  - Other Variables Included in the Regressions: Per capita non export income, export ratio
  - Goodness of Fit: 25 to 50%
  - Countries Covered: 50 developing countries
  - Time Period: 1953-55 and 1966-68

- Chelliah, Baas and Kelly (1975)
  - Significant Variables (Sign): Mining share (+), agriculture share (-)
  - Other Variables Included in the Regressions: Trade share, non mineral exports, per capita non export income
  - Goodness of Fit: 11 to 45%
  - Countries Covered: 47 developing countries
  - Time Period: 1969-71

- Tait, Grätz and Eichengreen (1979)
  - Significant Variables (Sign): Mining share (+), non mineral export share (+), export share (+)
  - Other Variables Included in the Regressions: Per capita income, per capita non export income, agriculture share
  - Goodness of Fit: 26 to 54%
  - Countries Covered: 47 developing countries
  - Time Period: 1972-76

- Tanzi (1981)
  - Significant Variables (Sign): Mining share (+), non mineral export share (+)
  - Other Variables Included in the Regressions: Per capita non export income
  - Goodness of Fit: 15 to 52%
  - Countries Covered: 34 Sub Saharan African countries
  - Time Period: 1977

- Tanzi (1992)
  - Significant Variables (Sign): Agriculture share (-), import share(+), foreign debt share (+)
  - Other Variables Included in the Regressions: Per capita income,
  - Goodness of Fit: 54%
  - Countries Covered: 88 developing countries
  - Time Period: 1978 -88

- Leuthold (1991)
  - Significant Variables (Sign): Trade share (+), agriculture share (-)
  - Other Variables Included in the Regressions: Foreign grants, mining share
  - Goodness of Fit: 38%
  - Countries Covered: 8 African countries
  - Time Period: 1973-81

- Stotsky and WoldeMariam (1997)
  - Significant Variables (Sign): Agriculture share (-), mining share (-), export share (+), per capita GDP (+), IMF dummy (+)
  - Other Variables Included in the Regressions: Manufacturing share, import share
  - Goodness of Fit: 57 to 94%
  - Countries Covered: 46 Sub Saharan African countries
  - Time Period: 1990-95

- Ghura (1998)
  - Significant Variables (Sign): Per capita income (+), agriculture share (-), trade openness (+), existence of oil and non oil mining sector (+), structural reforms (+), human capital development (+), inflation (-), corruption (-)
  - Other Variables Included in the Regressions: Percentage change in terms of trade, percentage change in real exchange rate, change in external debt to GDP ratio
  - Goodness of Fit: Not Reported
  - Countries Covered: 39 Sub Saharan African countries
  - Time Period: 1985-96

- Piancastelli (2001)
  - Significant Variables (Sign) [first specification]: Trade share (+), agriculture share (-), manufacturing share (+), services share (+)
  - Other Variables Included in the Regressions: Per capita GDP
  - Goodness of Fit: 38 to 84%
  - Countries Covered: 75 countries
  - Time Period: 1985-95
  - Significant Variables (Sign) [second specification]: Per capita GDP (+), mining share (-)
  - Other Variables Included in the Regressions: Import share, export share, manufacturing share, agriculture share, outstanding foreign debt
  - Goodness of Fit: 50%
  - Countries Covered: 6 oil producing Arab countries

- Eltony (2002)
  - Significant Variables (Sign): Per capita GDP (+), import (+), export (+), mining share (+), agriculture share (-), outstanding foreign debt (+)
  - Other Variables Included in the Regressions: Export share, manufacturing share
  - Goodness of Fit: 78%
  - Countries Covered: 10 non oil producing Arab countries
  - Time Period: 1994-2000

- Bird, Martinez-Vasquez & Torgler (2004)
  - Significant Variables (Sign): Population growth (-), agriculture share (-), inequality (-), shadow economy (-), institutions (+), entry regulations (-)
  - Other Variables Included in the Regressions: Per capita GDP
  - Goodness of Fit: 48 to 85%
  - Countries Covered: 110 developing and transitional countries
  - Time Period: 1990-99

### Key patterns across studies
- Recurring positive correlates of tax revenue ratios:
  - Per capita GNP/Per capita GDP (+)
  - Trade or export shares (+)
  - Mining or oil sector presence (+) in many specifications
  - Structural reforms and human capital development (+) in selected studies
- Recurring negative correlates of tax revenue ratios:
  - Agriculture share (-)
  - Inflation (-) and corruption (-) in some studies
  - Population growth (-), inequality (-), and shadow economy (-) in cross-country micro-studies
- Goodness of fit varies widely across studies, with reported ranges from 10 to 60% up to 57 to 94%, and some instances of a single reported value such as 54% or 78%

### Notes
- a. Dependent variable is ratio of tax revenue to GNP.
- b. Dependent variable is ratio of tax revenue (excluding social security payments) to GNP.
- c. Dependent variable is ratio of tax revenue to GDP.
- d. Dependent variable is ratio of tax revenue to GDP and ratio of current revenue (minus grants) to GDP.

*Source: Appendix D. Summary of Findings of Empirical Studies*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2007/_wp07184.pdf_